LabVenture

Swap

Comparing responses

Here we examine the difference in students responses when the order of activities within the LabVenture program are swapped.

• Normal order: Lobster, Otolith, Dissection, SST tracing
• Swapped order: SST tracing, Lobster, Otolith, Dissection

Though there are apparent differences in favored activities when the order of activies is swapped, there is no statistically significant difference. To assess this difference, I ran a Welch’s t.test to compare the difference in the average percentage of students ranking each activities. The results of the significance testing are in the table below.

Response options	Normal program	Swap	Method	P.Value	Significant
fav_plank	0.06142857	0.04375	Welch Two Sample t-test	0.50886320	False
fav_lobsters	0.68571429	0.76875	Welch Two Sample t-test	0.09302415	False
fave_dissection	0.10428571	0.07000	Welch Two Sample t-test	0.28089728	False
fave_tank	0.12285714	0.09375	Welch Two Sample t-test	0.31696407	False
fave_trace	0.02714286	0.01750	Welch Two Sample t-test	0.48773346	False
worst_plank	0.21285714	0.21625	Welch Two Sample t-test	0.94023854	False
worst_lob	0.04285714	0.03875	Welch Two Sample t-test	0.83798055	False
worst_dissection	0.20285714	0.18750	Welch Two Sample t-test	0.65027118	False
worst_tank	0.08428571	0.08750	Welch Two Sample t-test	0.89227730	False
worst_trace	0.45571429	0.46750	Welch Two Sample t-test	0.83383656	False
longer	0.32285714	0.30625	Welch Two Sample t-test	0.75861991	False
shorter	0.20285714	0.29250	Welch Two Sample t-test	0.11718421	False
same	0.44857143	0.38500	Welch Two Sample t-test	0.30005822	False

I ran the same statistical test on the average number of otolith/sst worksheets that were blank or covered in scribbles. Like the previous analysis, the change in activity order doesn’t seem to significantly impact engagement.

Worksheet	Normal program	Swap	Method	P.Value	Significant
otolith	1.444444	1.666667	Welch Two Sample t-test	0.7443417	FALSE
sst	1.000000	1.444444	Welch Two Sample t-test	0.3630250	FALSE

My next step is to assess if there are any differences in the multichoice responses, though my initial thoughts are that these will follow the same pattern as the clicker questions.

I’m not entirely sure what statistical test to do here with such a small sample size. However, I suspect that the differences in responses is not impacted by the order of activities.

General conclusions

Based on the descriptive data of the normal visits and the stats run on the swap data, it seems unlikely that the order of activities does not impact engagement. However, this is not necessarily a bad thing! There is a clear pattern of understanding between what is happening in the Gulf of Maine, how we know that it’s happening, and why it’s important. The fact that these conclusions can still be drawn regardless of activity order is significant. Statistics is all about interpretation

You may remember from stats courses the concept of the null and alternate hypothesis. The null assumes no relationship between the variables. The 0.05 alpha is what we chose as our significant value. The Welch’s t-test for all the different variables yielding p-values larger than the alpha accepts this null hypothesis.

There are other types of analyses we could run on this info (if we had more time). One thought is to do an adjacency analysis to see which responses to the what is happening in the Gulf of Maine questions are connected to which how and why responses. But I think in the interest of time prior to the conference, we’ll need to stick to these descriptive stats.